Beruflich Dokumente
Kultur Dokumente
Never Stand Still Faculty of Engineering School of Civil and Environmental Engineering
Vehicle Stability Testing for Flood Flows
May 2017
by
G P Smith, B D Modra, T A Tucker and R J Cox
Project Details
Client Name NSW State Emergency Service, NSW Office of Environment and Heritage
Client Address
Client Contact SES, Belinda Davies belinda.davies@ses.nsw.gov.au
OEH, Duncan Mcluckie duncan.mcluckie@environment.nsw.gov.au
Client Reference
Document Status
This report was produced by the Water Research Laboratory, School of Civil and Environmental Engineering,
University of New South Wales for use by the client in accordance with the terms of the contract.
Information published in this report is available for release only with the permission of the Director, Water
Research Laboratory and the client. It is the responsibility of the reader to verify the currency of the
version number of this report. All subsequent releases will be made directly to the client.
The Water Research Laboratory shall not assume any responsibility or liability whatsoever to any third
party arising out of any use or reliance on the content of this report.
Executive Summary
Every year floods cause enormous damage and loss of life on a global scale. An analysis of
global statistics for loss of life showed that inland floods (river floods, flash floods and drainage
floods) caused 175,000 fatalities and affected more than 2.2 billion people between 1975 and
2002 (Jonkman, 2005). More recent global analysis noted that 59,092 flood fatalities occurred
worldwide between 2005 and 2014 (International Federation of Red Cross and Red Crescent
Societies, 2015).
In a recent detailed analysis of flood fatalities in Australia, Haynes et al. (2016) noted that 1,859
people have died in floods since 1900 and that 178 of these flooding related deaths have
occurred since 2000. The study noted that while flood fatality rates are generally falling per
capita, the number of fatalities that occurred in vehicles, particularly four wheel drive (4WD)
vehicles has increased in the last fifteen years.
Beyond the unfortunate occurrence of flood related fatalities, an enormous amount of time and
resources are invested by emergency response organisations rescuing people who have entered
floodwaters. Information sourced from the NSW SES website to the end of September 2016
indicates that nearly 550 flood rescues have been performed by NSW SES in 2016 alone (NSW
SES, 2016). During the flooding of June 2016, NSW SES performed 300 flood rescues,
approximately a third of which involved rescuing people from flooded vehicles (NSW SES, 2016).
It is clear that there is a need to better understand and quantify the mechanisms by which
vehicles can become unstable in floodwaters. The research and analysis described in this report
aims to improve the knowledge and information available to describe the vulnerability of vehicles
as they enter floodwaters and to quantify the flow conditions that might cause vehicles to
become vulnerable to being washed away. This knowledge can be used to better inform
emergency managers and floodplain managers seeking to plan for, and respond to, flood
emergencies.
This report describes an investigation of vehicle stability recently conducted at the UNSW Water
Research Laboratory of the School of Civil and Environmental Engineering, UNSW Sydney. The
report includes:
A comprehensive literature review of pre-existing research and anecdotal evidence
describing the vulnerability of vehicles in floodwaters;
A description of the research scope;
A description of the test methodology and test results;
Discussion of the measured data;
Interpretation of the data for use by decision makers in floodplain management and
emergency planning.
Vehicle stability testing incorporated measurement of full prototype scale traction and buoyancy
forces for threshold of movement of a vehicle in various depths of water in WRLs tow tank, and
hydrodynamic force measurement on a 1:18 scale model vehicle to determine the equivalent
hydrodynamic flow forces to reproduce the measured, full scale threshold of movement forces.
The traction force tests are considered novel in research on the stability of vehicles, and show
that traction decreases rapidly with floodwater depths above the vehicles floor pan level. The
testing confirmed that the rear wheels (wheels not weighted by the engine) break traction far
earlier than the front wheels.
-i-
Hydrodynamic testing showed that coefficients of drag for a vehicle in free surface water flows,
perpendicular to the vehicle, are far higher than typical wind tunnel based aerodynamic drag
coefficients. Flooded vehicle drag coefficients are consistent with the drag coefficients predicted
by Hoerner (1965) for water surface piercing drag and are complicated by the vehicle geometry
and low depth to vehicle length ratios. The coefficient of drag for a flooded vehicle varies non-
linearly with the flow condition, but is generally within the range 1.2 and 2.0 for typical,
dangerous flood conditions.
The test program and analysis presents a novel approach to the derivation of stability curves for
vehicles in flood flows, by dissecting the forces that impact the vehicle, applying appropriate
factors of safety for real-life flood conditions (as opposed to controlled laboratory conditions),
and evaluating vehicle stability directly from the stability balance equation.
Observations from the testing confirm that where floodwaters impact the body of the vehicle, the
vehicle is at greater risk of losing traction and being washed downstream. Flows that do not
impact the body of the vehicle (hitting the wheels only) are generally considered safer for that
vehicle.
Analysis of raw results from this investigation showed that the subject test vehicles were more
stable in flood flows than the flood hazard vulnerability thresholds typically applied for floodplain
management and emergency management as documented by guidelines such as Australian
Rainfall and Runoff Review Project 10 (ARR P10) by Shand et al. (2011) and AEMI Handbook 7
(2014). However, mitigating factors described in this report, notably i) that the tested vehicles,
while representative, were not the smallest and therefore the most vulnerable in their vehicle
class and ii) that controlled laboratory conditions need to be interpreted to uncontrolled real life
flood conditions meaning that a level of conservatism on this studys results is required. On this
basis, the interim flood hazard threshold curves from ARR P10 summarised below for the two
vehicle classes tested in this investigation are confirmed as appropriate for floodplain
management and emergency planning.
Limiting
Kerb Ground Limiting High
Class of Length Limiting Equation of
Weight Clearance Still Water Velocity
Vehicle (m) Velocity3 Stability
(kg) (m) Depth1 Flow
Depth2
Small
< 4.3 < 1250 < 0.12 0.3 0.15 3.0 D.V 0.3
passenger
Large 4WD > 4.5 > 2000 > 0.22 0.5 0.3 3.0 D.V 0.6
1 2 3
At velocity = 0 m/s; At velocity = 3 m/s; At low depth
Regardless of these findings, it should be noted that estimation of the depth and velocity
characteristics of floodwaters over a roadway in a real life flood emergency situations is fraught
with difficulty. Floodwaters are often sediment laden and murky, making it difficult to assess the
flow depth. In many cases, the flows overtopping flooded bridges or roadways may have been
washed out below the floodwaters making the roadway unpassable. Without sophisticated
equipment, flow velocities are notoriously difficult to assess, even to the trained eye. For these
- ii -
reasons, the ad-hoc use of the above tabulated stability criteria by individuals in real flood
situations to assess the safety of a road crossing is actively discouraged.
This report does not account for behaviour of drivers in flood emergencies, which remains a
substantial factor in the ongoing loss of life in vehicles in floodwaters.
Having gained substantial knowledge of the vulnerability of vehicles exposed to floodwater, the
authors of this report support the best advice of the Australian State Emergency Services to
individual drivers which remains Never drive, ride or walk through floodwater.
- iii -
Contents
Executive Summary i
1. Introduction 1
2. Background: Vehicle Stability in Floodwaters 2
2.1 Vehicle stability information literature review 3
2.1.1 Anecdotal evidence 9
2.1.2 Summary findings 10
3. Research Scope 11
4. Methodology 12
4.1 Overview 12
4.2 Formulation of the problem 14
4.2.1 Friction force of a vehicle 14
4.2.2 Hydrodynamic drag on a vehicle 16
5. Full Scale Vehicle Traction Tests 21
5.1 Overview 21
5.2 Test vehicles 21
5.3 Tyre coefficient of friction 23
5.4 Overview of traction tests 24
5.5 Test results 26
6. Hydrodynamic Force Testing (1:18 Scale) 32
6.1 Overview 32
6.2 Subcritical flows (tailwater controlled) 34
6.3 Critical flows (broad crested weir) 34
6.4 Supercritical flow (undershot gate) 35
6.5 Horizontal force testing results 36
7. Discussion of Results 39
7.1 Observations 39
7.2 Comparison of tested vehicles to other vehicles 41
7.3 Testing uncertainties 43
7.4 Interpretation of test information for floodplain management and emergency
management 43
7.4.1 Preamble 43
7.4.2 Comparison with existing flood hazard curves 44
8. Conclusions 49
8.1 Recommendations 49
9. Acknowledgements 51
10. References 52
-i-
List of Tables
List of Figures
Figure 2-1: Combined flood hazard curves (after Smith et al., 2014; AEMI, 2014) 2
Figure 2-2: Vehicle stability failure mechanisms (after Shand et al., 2011) 3
Figure 2-3: Car floating in deep, slow flow
https://www.youtube.com/watch?v=LC5ld79joIA (accessed 23 March
2017) 4
Figure 2-4: Car washed down an urban street in shallow, fast flow
https://www.youtube.com/watch?v=3HrdgaiM9sY (accessed 23 March
2017) 5
Figure 2-5: 4WD utility washed off a causeway in Australia
https://www.youtube.com/watch?v=Va8w7Jng9rM (accessed 22 March
2017) 5
Figure 2-6: Representation of all the results from literature (instability points and limit
functions) obtained in experimental and theoretical studies (after
Martinez-Gomariz et al., 2016) 8
Figure 2-7: Youtube footage of flooded cars in the Toowoomba floods of January 2010
(https://www.youtube.com/watch?v=kYUpkPTcqPY accessed 20 January
2017) 10
Figure 4-1: Schematic illustration of prototype scale test rig 12
Figure 4-2: Nissan Patrol in WRLs wave basin facility, preparing for a winch test from
the rear axle 13
Figure 4-3: Slip curves for various road surfaces, extracted from Gerard (2006) 16
Figure 4-4 Drag coefficient of surface-piercing circular cylinders (on area d times h)
as a function of Froude number, extracted from Hoerner (1965) 18
Figure 4-5 Drag coefficient of surfacepiercing flat plates (on area b times h),
extracted from Hoerner (1965) 19
Figure 4-6: Hoerners (1965) formulation for surfacing piercing drag, extracted from
Bowen et. al. (2011) 19
Figure 4-7: Flood flows over a causeway approaching Froude Fr = 1 20
Figure 5-1: Toyota Yaris in WRLs test facility, preparing for a winch test from both
axles 22
- ii -
Figure 5-2: Nissan Patrol in WRLs wave facility, preparing for a winch test from the
rear axle 22
Figure 5-3: Winch assembly in WRL Lab#2, including (a) framework, winch and
dynamometer, (b) pulley and wheel attachment 23
Figure 5-4: Rapid filling of WRLs test tank (wave basin). The winch assembly is also
shown. 25
Figure 5-5: Toyota Yaris in WRLs wave basin facility, in 0.6 m water depth with rear
wheels clear of the floor 26
Figure 5-6 The lead author demonstrating zero traction at the rear wheels for the
Nissan Patrol 4WD in WRLs Lab#2 testing tank. Water depth is
approximately 1.0m. 26
Figure 5-7: Traction force time series for tests on the Toyota Yaris with one driver
occupant, winched from the rear axle only 27
Figure 5-8: Traction force time series for tests on the Toyota Yaris with four
occupants, winched from the rear axle only 28
Figure 5-9: Traction force time series for tests on the Toyota Yaris with one driver
occupant, winched from both axles 28
Figure 5-10 Traction force time series for tests on the Nissan Patrol with one driver
occupant, winched from the rear axle 29
Figure 5-11: Measured peak horizontal traction force for a Toyota Yaris 2006 in varying
static water levels. Tests were conducted on three scenarios with varying
passenger loads and tow points 30
Figure 5-12: Measured peak horizontal traction force for a Nissan Patrol 4WD 1998,
winched from the rear axle only, with one occupant 31
Figure 6-1: Test assembly for testing of the horizontal hydrodynamic forces. Flow is
from the left of the picture 33
Figure 6-2: Testing of the model in subcritical flows: Fr = 0.3, d= 0.71m, v =
0.80m/s (prototype) 34
Figure 6-3 Testing of the model on a broad crested weir 35
Figure 6-4: Testing of the model in supercritical flows d = 0.225m, v=4.0m/s
(prototype), CD = 1.86, Fr = 2.66 36
Figure 6-5 Coefficient of Drag for a Toyota Yaris based on the horizontal
hydrodynamic force tests 38
Figure 7-1: Vehicle stability deducted from prototype scale measurements rear
wheel horizontal force thresholds versus water depth 39
Figure 7-2: Comparison of the Force vs Depth relationship for the Hydrodynamic Tests
and the Winch Tests; Toyota Yaris 41
Figure 7-3: Stability Curve for a Toyota Yaris based on full scale traction tests, upper
limits for the Coefficient of Drag CD = 2.0 based on scale testing, and
points scaled to a flood condition Coefficient of Friction of 0.3 45
Figure 7-4: Stability Curve for a Nissan Patrol based on full scale traction tests, upper
limits for the Coefficient of Drag CD = 2.0 based on scale testing, and
points scaled to a flood condition Coefficient of Friction of 0.3 46
- iii -
1. Introduction
Every year floods cause enormous damage and loss of life on a global scale. An analysis of
global statistics for loss of life showed that inland floods (river floods, flash floods and drainage
floods) caused 175,000 fatalities and affected more than 2.2 billion people between 1975 and
2002 (Jonkman, 2005). More recent global analysis noted that 59,092 flood fatalities occurred
worldwide between 2005 and 2014 (International Federation of Red Cross and Red Crescent
Societies, 2015).
In a recent detailed analysis of flood fatalities in Australia, Haynes et al., (2016) noted that
1,859 people have died in floods since 1900 and that 178 of these flooding related deaths have
occurred since 2000. The study noted that while flood fatality rates are generally falling per
capita, the number of fatalities that occurred in vehicles, particularly four wheel drive (4WD)
vehicles has increased in the last fifteen years.
That the greatest proportion of flood related fatalities occurred while the person was
attempting to cross a bridge causeway or road, whether the person was in a vehicle or
on foot;
Most victims were capable of independent action (not impaired in any way e.g. by
alcohol consumption) and aware of the flood, however the speed and depth of the water
took them by surprise.
Beyond the unfortunate occurrence of flood related fatalities, an enormous amount of time and
resources are invested by emergency response organisations rescuing people who have entered
floodwaters. Information sourced from the NSW SES website to the end of September 2016
indicates that nearly 550 flood rescues have been performed by NSW SES in 2016 alone (NSW
SES, 2016). During the flooding of June 2016, NSW SES performed 300 flood rescues,
approximately a third of which involved rescuing people from flooded vehicles (NSW SES, 2016).
It is clear from this brief introduction that there is a need to understand the mechanisms by
which vehicles become unstable in floodwaters. The research and analysis described in this
report aims to improve the knowledge and information available, to describe the vulnerability of
vehicles as they enter floodwaters and to quantify the flow conditions that might cause vehicles
to become vulnerable in floodwaters. This knowledge can be used to better inform emergency
managers and floodplain managers seeking to plan for, and respond to, flood emergencies.
Figure 2-1: Combined flood hazard curves (after Smith et al., 2014; AEMI, 2014)
Vehicle stability thresholds expressed in the form illustrated in Figure 2-1 or similar can be
applied in various ways. In a generic sense, the information can be applied to support decisions
where vehicular access to a potentially flooded area might need to be controlled or restricted.
For example, an emergency manager might use the thresholds to classify available flood
1
Flood Hazard: Potential loss of life, injury and economic loss caused by future flood events. The degree of hazard varies
with the severity of flooding and is affected by flood behaviour (extent, depth, velocity, isolation, rate of rise of floodwaters,
duration), topography and emergency management.
This report section summarises the ARR P10 review, but has been extended with references
published in the intervening period.
There are essentially three reasons why vehicles are vulnerable in floodwaters:
Though many 4WD vehicles are designed to be driven in water, the reality is that
for many vehicles, entering water causes the engine to stop, preventing the
vehicle from being driven further into, or out of the water by the driver;
The vehicle floats; and
The vehicle becomes difficult to control.
The safety of people during flood events can be compromised when vehicles they are travelling
in are exposed to flood flows which cause the vehicle to become unstable by losing traction
(frictional instability through sliding), to topple because of uneven surfaces or to become
buoyant (floating). These mechanisms can lead to a complete loss of control over the vehicle
resulting in the vehicle being swept downstream (Shand et al., 2011; Haynes et al., 2016).
These stability failure mechanisms are illustrated in Figure 2-2.
Figure 2-2: Vehicle stability failure mechanisms (after Shand et al., 2011)
Once the flow thresholds for vehicle stability have been exceeded (Figure 2-1) and the vehicle
becomes unstable, the vehicle will typically be moved by the force of the water in the direction
of flow and observations of video footage of actual floods demonstrate that the vehicle often
then drifts under the influence of the floodwaters into deeper, faster moving, more dangerous
flow. Figures 2-3 to 2-5 (and the linked videos) illustrate situations where vehicles floated in
deep subcritical flood waters or were swept away by supercritical flood waters with high flow
velocity and relatively low flow depth.
The weight distribution in most vehicles is uneven with a greater proportion of the weight
located near the engine. In most vehicles, this means that the rear of the vehicle floats first (as
illustrated in Figure 2-2) with the front wheels below the engine remaining on the ground for
longer. Where floodwaters have adequate velocity, the rear of the vehicle is prone to swing
around, pivoting on the front wheels until the vehicle is aligned with the flow direction.
More recent investigations of vehicle stability are not included in the review by Shand et al.,
(2011). Recent experimental investigations have been undertaken using scale model die-cast
vehicles in a horizontal hydraulic flume at the Hydro-environmental Research Centre of Cardiff
University, United Kingdom (Shu et al., 2011; Xia et al., 2011). These studies evaluated the
theoretical forces exerted on a fully submerged and static vehicle under flow conditions to derive
an expression for the instability threshold. The drag and frictional coefficients were included
implicitly within the expression and derived by laboratory flume testing using model vehicles.
This methodology differed from previous studies, where reaction forces were measured for a
geometrically scaled vehicle subject to flows and the instability threshold was calculated for
various coefficients of friction.
Xia et al., (2011) tested three types of vehicle using an initial geometric scale of 1:43 with
verification tests at a larger 1:18 scale. Shu et al. (2011) tested the same three vehicle types
namely Mitsubishi Pajero, BMW M5, Mini Cooper, undertaking all tests at 1:18 scale. Both
testing programs assumed all wheels to be locked and the vehicle orientated parallel to the flow
direction. Velocity was increased for a given depth until vehicle motion was initiated.
Experimental results were used to calibrate the parameters within the theoretical expression for
instability thresholds.
The studies differed in that Xia et al., (2011) did not correctly scale the vehicle density, and
therefore mass of the model vehicles, according to dynamic similarity principles but rather used
a relative density term in the analytical solution to adjust the measured data to match the
prototype. The resultant values of Xia et al. (2011) differed significantly from other studies (by
up to an order of magnitude) with vehicles becoming submerged before moving and were
inconsistent with qualitative evidence from the field (i.e. during the Queensland floods in
February 2011 (https://www.youtube.com/watch?v=kYUpkPTcqPY accessed 24 March 2017) and
the Japan Tsunami in March 2011 (https://www.youtube.com/watch?v=7trfmf5SuQs accessed
24 March 2017) where cars floated well before they became submerged.
Shu et al., (2011) correctly scaled the density and mass of model vehicles and obtained D.V
values slightly higher than those found in earlier investigations. Shu et al., (2011) had the
model test vehicles filled with foam to resist ingress of water. This was thought to better
represent modern vehicles with significantly improved dust seals than the older cars tested in
most other investigations. Shu et al., (2011) determined the friction coefficients to range from
0.39 to 0.68 for their models, quoted to be within prototype ranges given by Gerard (2006).
The differences in stability thresholds noted in Shu et al., (2011) are likely due to the higher
coefficient of friction in the Shu et al., (2011) models ( = 0.39 to 0.68) compared to the lower
friction coefficients ( = 0.3) adopted by earlier studies (e.g. Bonham and Hattersley, 1967;
Keller and Mitsch, 1992). Sensitivity testing of Shu et al., (2011) using a reducing frictional
Shu et al., (2011) verified their results against two field observations of cars moving during flood
events. The threshold of vehicle motion and the flow rate, which may be considered safe for
vehicle passage are likely to be lower than values observed for vehicles already moving in flood
flow.
Toda et al., (2013) also completed scale model testing of a sedan at 1:10 scale and a minivan at
1:18 scale in a 1 m wide flume. Froude similarity principles were used for scaling. Considerable
effort was invested in adequately representing the weight distribution between front and rear of
the vehicles and buoyancy of the scale models. Models were tested with the vehicle oriented so
the vehicle pointed into the direction of flow (0 degree) and oriented across the flow (90 degree)
to determine instability. Friction coefficients were measured at both 0 and 90 degree orientation
to the flow with friction values of 0.26 and 0.57, respectively, for the sedan and friction
coefficients of 0.42 and 0.65 for the minivan. The authors provided a simplified conclusion that
cars would move in floodwaters greater than 0.5 m deep and flowing at a velocity greater than
2 m/s.
The criteria proposed by Shand et al., (2011) presented in Table 2-1 have taken into
consideration the safety criteria for people (Cox et al., 2010), ensuring that, in event of vehicle
failure, safety was not compromised once people abandoned their cars.
Table 2-1: Proposed draft interim criteria for stationary vehicle stability
(after Shand et al., 2011)
Limiting
Kerb Ground Limiting High
Class of Length Limiting Equation of
Weight Clearance Still Water Velocity
Vehicle (m) Velocity3 Stability
(kg) (m) Depth1 Flow
Depth2
Small
< 4.3 < 1250 < 0.12 0.3 0.1 3.0 D.V 0.3
passenger
Large
> 4.3 > 1250 > 0.12 0.4 0.15 3.0 D.V 0.45
passenger
Large 4WD > 4.5 > 2000 > 0.22 0.5 0.2 3.0 D.V 0.6
1 2 3
At velocity = 0 m/s; At velocity = 3 m/s; At low depth
Many of the studies listed above, particularly Shand et al., (2011), Xia et al., (2011) and Shu et
al., (2011) recommended full scale testing to investigate thresholds of flooded vehicles under
real and more complex circumstances.
The internet, particularly web sites such as youtube.com, provides a wealth of documented
anecdotal observations of vehicles in floodwaters. Internet video footage sampled as part of this
research generally validates the conclusion of Haynes et al., (2016) that vehicles most
commonly come into contact with floodwaters when they are deliberately driven into
floodwaters, either at a bridge or causeway crossing.
Observations of vehicles in these circumstances show that most commonly the vehicle enters the
flood waters with flow passing perpendicular to the vehicle direction, i.e. the road or causeway
that is being overtopped by floodwater is more or less perpendicular to the flow direction. This
is generally the worst orientation for vehicle stability as it provides the greatest vehicular cross-
sectional area to the flow, and therefore the highest possible hydrodynamic flow force is applied
to the vehicle.
Video footage also demonstrates, that in this scenario, the most common mode of failure is for
the vehicles rear wheels to lose traction causing the vehicle to rotate nose first into the flow and
in many cases then be washed off the roadway into deeper water as illustrated in Figure 2-7.
This mode of failure is reasonably anticipated since the typical configuration of modern vehicles
is with a forward mounted engine, in front of a near water-tight cabin and boot/trunk rearwards
of the rear axle. As the vehicle becomes inundated, the rear of the vehicle is more buoyant than
the front with less downwards force due to the vehicle weight at the rear wheels. Less weight
over the rear axle means less friction available from the tyres, and so the rear wheels break
traction first.
If the rear wheels break traction in moving floodwaters, the vehicle has a tendency to pivot on
the front wheels eventually pointing the front of the vehicle into the direction of flow. If the
vehicle remains on a reasonably flat surface, then with water pressure sometimes rising over the
bonnet of the vehicle, the vehicle may become more stable and stop moving. However, in many
recorded instances the vehicle rolls off the road over the edge of a causeway or bridge, or may
simply roll away downstream. For this reason, the point of instability of the vehicle is defined in
this study as the point at which any axle (two wheels) loses traction and the vehicle rotates or
translates sideways.
The review of currently available literature concluded that that there are a number of limitations
to the previous research on vehicle stability in floodwaters, including:
Little previous work has aimed to separate and understand the various forces that act on a
vehicle in floodwaters. This limits the ability of the scientific community to study specific aspects
of vehicle stability in detail, such as buoyant behaviour, tyre-road friction or hydrodynamic
forces. This limits the ability to determine uncertainty in testing programs, or develop suitable
factors of safety, and to recommend defensible parameters to use in floodplain management or
emergency management and operations when considering vehicular access in flood prone areas.
The research scope for this project was driven by the need to better understand and quantify the
mechanisms by which vehicles become unstable in floodwaters. By better understanding the
stability limits of vehicles in flood flows, and quantifying the flow conditions that might cause
vehicles to become vulnerable to being washed away by floodwaters, better informed emergency
managers and floodplain managers can more effectively plan for, and respond to, flood
emergencies.
The research and analysis described in this report aims to improve the knowledge and
information available. The study aims were to:
1. Determine the likely conditions in which a vehicle may be washed away in floods;
2. Determine the buoyant behaviour of vehicles and the subsequent impact on road-tyre
friction by direct measurement of a sample of full scale vehicles in water;
3. Determine the hydrodynamic forces on a vehicle and the vehicles coefficient of friction
by scale physical modelling;
5. Based on the results of testing the friction and hydrodynamic forces, as well as other
available literature, determine suitable stability curves for passenger vehicles and large
4WD vehicles.
Direction for the research scope was also derived from the key findings of Haynes et al. (2016)
that:
The greatest proportion of flood related fatalities occurred while the person was
attempting to cross a bridge, causeway or road, whether the person was in a vehicle or
on foot;
Most victims were capable of independent action (not impaired in any way e.g. by
alcohol consumption) and aware of the flood, however the speed and depth of the water
took them by surprise.
The following research scope was proposed to address these primary research needs:
i. Measure the stability of vehicles when exposed to floodwaters at prototype (full) scale
using real vehicles;
a. Devise tests that are representative of typical scenarios where vehicles on
bridges or causeways have been washed away in floodwaters;
ii. Based on these measurements, review the flood hazard vulnerability criteria for flood
exposed vehicles.
4.1 Overview
The overriding limitation with currently available research of vehicle stability is that all previous
research assessment has been conducted at model scale. A particularly novel aspect of this
current research is the world first, direct measurement of the forces required to cause a
passenger vehicle to become unstable in floodwaters at full scale.
The test methodology was designed to separate the response of the vehicle exposed to
floodwaters to both friction and hydrodynamic forces. Two rounds of testing were completed:
a) Full scale measurement of the force required to overcome the friction force of a vehicles
tyres on the roadway surface in various water depths;
b) Knowing the full scale forces required to initiate vehicle instability, scale model testing of
the hydrodynamic forces to reproduce the vehicle instability.
Full scale testing of moving flows around the subject vehicle could not be employed in this study
due to the very large flows (upwards of 12 m3/s) required to achieve it. Instead a number of
tests were conducted by winching the test vehicle relative to a water body at various depths in a
tow tank. This allowed full scale testing of buoyant friction forces under varying water levels.
Moving of the full scale vehicle to determine the vehicle instability force was achieved using a
high powered winch to drag the vehicle sideways as illustrated schematically in Figure 4-1 and in
practice in Figure 4-2.
Forces were considered separately at the vehicle front and rear axles. This allowed
consideration of stability of the vehicle as a whole, or stability failure when traction is lost at one
axle first and then rotated into alignment with the direction of flow. Specifically, a substantial
part of the testing focused on the stability of the rear axle as floating of the rear of the vehicle
has previously been identified as the most likely initiation point of instability of a vehicle exposed
to floodwaters.
The choice of the axles as reference points on the vehicle provides a number of advantages.
Fundamentally it allows consideration of the forces through their point of action (i.e. friction at
the wheels) and particularly the separation of forces at the front and rear axles allows the
measurement of the instant of traction loss at the rear axle. The axle provides a convenient
location for vehicle force measurement that does not load or unload the vehicle suspension..
Scale model testing was then completed to determine the equivalent hydrodynamic forces
required to reproduce the prototype scale vehicle instability forces. Further, detailed description
of the experiment methodology is provided in the following sections.
FH > FF (1)
The friction force is dependent on the downward force on the wheels, and may be considered as
a vehicle as a whole, at an individual wheel, or at an axle:
FF = (W B L) (2)
Where:
W is the weight of the vehicle;
B is the buoyancy which is dependent on the water depth (d);
L is the uplift force due to water impacting the vehicle which is dependent on the water depth
and velocity; and
is the coefficient of friction between the tyres and the road.
Note that once the tyres are sliding, the coefficient of force () decreases and it becomes more
difficult for the vehicle to regain traction.
The hydrodynamic force can be expressed in the form of a drag force as:
FH =0.5..A.CD.v2 (3)
Where:
is the water density;
A is the projected (profile) area of the vehicle impacted by water;
CD is the coefficient of drag which is dependent on the water velocity and depth; and
v is the depth-averaged flow velocity.
The coefficient of drag is a complex term capturing the influence of the shape of the vehicle and
the response of the flow to the obstruction. The coefficient of drag needs to be determined
experimentally for each vehicle, and varies with flow depth and velocity.
Combining equations (1), (2) and (3), including the dependencies, a vehicle is stable when:
In determining the conditions for vehicle stability, the focus of this research is the investigation
of factors that affect the friction force ( and B), and the coefficient of drag for the vehicle. A
discussion of the theoretical background and relevant knowledge on each of these parameters is
provided in the following sections.
Bonham and Hattersly (1967), after a detailed review of studies completed between 1930-1936
at the Road Research Laboratory in England (Bird et al., 1936) and consultation with Olympic
Tyre and Rubber Co., arrived at a single conservative value for the coefficient of tyre friction.
This was based on a braking force coefficient of 0.5 on a worst case scenario of a smooth
causeway surface on a wet country road. Adjustments were made for a reduced sideways force
compared with braking force (10%), the reduced slipping force compared with peak braking
While this is a suitably conservative value where actual data is limited, it is worth revisiting the
range of friction coefficients likely to be encountered in real world conditions. This review
considers both changes in tyre and road technology and modern motoring conditions for the
interpretation of measured test results.
Most available tyre traction data relates to straight line traction (braking), or cornering at speed.
Limited analysis is available on the lateral grip of tyres while stationary. Rubber has a unique
property whereby the force required to initially break traction is higher than the force required to
maintain motion once traction is lost (Wong, 1993). In the available literature, these forces are
known as Peak and Sliding forces. It is important to distinguish between these forces for vehicle
stability, where it is unlikely that a vehicle will regain traction once it has been lost.
Wong (1993) provides typical coefficient of friction values for a range of conditions as
reproduced in Table 4-1. This information is derived from an investigation in 1957 (Taborek,
1957), so is somewhat dated, but broadly consistent with values for modern tyres and
conditions. Note that the values in Table 4-1 are for the coefficient of adhesion between rubber
and concrete, rather than the braking effort coefficient. The braking effort coefficient more
broadly includes the dynamics of the tyre at speed and under braking loads.
Table 4-1 Coefficient of Friction (adhesion) values between tyres and road surfaces, from Wong
(1993)
A number of important conclusions can be drawn from this information. Significantly, concrete
surfaces perform significantly better than asphalt in the wet by up to 40%. This alone is an
important factor in the stability of vehicles on bridges and causeways. The presence of dirt and
gravel on the roadway surface significantly decreases traction. Causeways will likely be affected
during floods as mud, sand, gravel and debris are washed downstream and over the causeway.
In particular, once vehicle sliding is initiated, the friction force reduces markedly (see sliding
factors in Table 4-1), meaning that it is likely that the vehicle will continue to move once initial
traction is lost.
There is a wealth of more recent information on tyre friction, largely targeted at vehicle accident
forensics. However the focus of this traction research is on braking deceleration as an average
over a distance, rather than the instantaneous peak traction, and is based on the braking effort
rather than coefficient of adhesion. As such many of the quoted values are lower than those
provided in Wong (1993) (see http://www.mfes.com/friction.html for many references).
Figure 4-3: Slip curves for various road surfaces, extracted from Gerard (2006)
Figure 4-3 presents the friction slip curves from Gerard (2006). This figure demonstrates that
there is a peak in the friction coefficient at low slip rates that is significantly larger than for a
fully slipping tyre. The peak is exaggerated for higher friction surfaces, but almost non-existent
for slippery surfaces.
In the traction tests conducted in the present investigation, the vehicle is towed at a near
constant rate set by the capability of the winch, which in turn varies with the load on the winch.
As the winch pulls the vehicle through the peak of the slip friction it slows down, then as the
vehicle begins to move and slip increases, the speed of the winch also increases. Thus a quasi-
balance is achieved on the falling part of the slip curve where load on the winch is balanced by
the friction of the tyres.
A detailed literature review found that there is extensive data on aerodynamic drag on vehicles
travelling forwards in air. However, there is very limited data in available literature on
hydrodynamic drag forces on a vehicle motoring in an open channel water flow (i.e. water with a
free surface) and even less information for the specific case associated with vehicle fatalities in
floods where the flows are perpendicular to the vehicle.
Early work by Hoerner (1965) provides detail on the components that contribute to the drag of a
partially submerged object analogous to a vehicle in flood water. The drag on an object in a flow
can be broken into two components:
1. Profile Drag: due to the displacement of flow around the object which is dependent on
the shape and cross-sectional area of the object; and
2. Viscous Drag (Skin Friction): caused by the interaction of the objects surface with the
viscous fluid, which is dependent on the length of the body and surface roughness.
3. Wave Drag: due to the deformation of the water surface, i.e. the generation of a wake;
and
4. Spray Drag: the upward jetting of water into the air at the object.
Hoerner (1965) investigated drag forces on bodies in free-surface flows in detail, and provided
drag coefficients for flat plates and cylinders piercing the surface of water flow in different flow
regimes (Figure 4-4 and Figure 4-5). The cylinder provided substantially less drag than a flat
plate, with a peak coefficient of drag for a short flat plate of approximately 1.7. This drag
coefficient is also consistent with typical drag coefficients for a flat plate with infinite length (CD
= 1.8).
The relative contributions of the three components of form drag for a surfacing piercing plate
was summarised in Bowen et. al. (2011), see Figure 4-6. This is for the specific case where the
depth of penetration into water is equal to the plate width, so the peak in CD is expected at a
Froude number Fr=1.
By way of background, the Froude number, Fr, is a dimensionless value that describes different
flow regimes of the open channel flow encountered on floodplains. The Froude number is a ratio
of inertial and gravitational forces of the flow and can be described by the equation:
= (5)
Where:
v = Depth-averaged flow velocity
D = Hydraulic depth (cross sectional area of flow / top width)
g = Gravity acceleration
When:
Fr = 1, critical flow,
Fr > 1, supercritical flow (fast rapid flow),
Fr < 1, subcritical flow (slow / tranquil flow)
Hoerners (1965) description of drag on a surface piercing object provides some important
insights for this study. A significant peak in the coefficient of drag on a flat plate occurs around
a Froude Fr=1, which may be around 50% higher than in other flow regimes. This peak moves
to higher Froude numbers as the depth to width ratio of the object decreases. This may be a
particularly important point of interest for the case of a vehicle passing over a bridge or
causeway on a floodplain where the change in floodway flow area as is conducive to producing
flood flow conditions approaching Froude Fr = 1 (see Figure 4-7). At the condition near Froude
Fr = 1, the free-surface becomes undular since small changes in flow energy lead to large
changes in flow depth, substantially increasing the drag on the object. This flow behaviour can
also be observed anecdotally in photos and videos of flooded vehicles crossing causeways and in
scale laboratory testing, with deep standing waves occurring at each end of the vehicle.
The plate scenarios tested by Hoerner (1965) are somewhat different to the case of a vehicle on
a road because:
The vehicle shape is significantly more complex including a void underneath the vehicle
above the road surface;
The situation of a vehicle in floodwaters is substantially more complex than the simple shapes
studied by Hoerner (1965). The vehicle may be exposed to shallow water flows leading to an
additional interaction with the lower (road surface) boundary. The width verses depth of the
vehicle is outside the range of conditions covered by Hoerner (1965) and the vehicle itself has a
complex shape, particularly with the presence of a second set of wheels behind the first, as well
as a relatively rough vehicle underbody.
The additional component of skin friction underneath the vehicle is acknowledged, but has not
been addressed as it is considered to be small compared to the form drag components of the
vehicle.
Figure 4-4 Drag coefficient of surface-piercing circular cylinders (on area d times h) as a function
of Froude number, extracted from Hoerner (1965)
Figure 4-6: Hoerners (1965) formulation for surfacing piercing drag, extracted from
Bowen et. al. (2011)
Uplift forces
Vertical forces on the vehicle due to hydrodynamic effects are referred to here as uplift forces,
though it has been observed that these may also be negative, pulling the vehicle downwards.
Discussions with the authors of Gordon and Stone (1973) suggest that uplift due to the impact of
flow into a vehicles wheel would contribute to destabilisation of the vehicle. But like horizontal
drag, the processes leading to a vertical uplift force are complex. Flows impacting the upstream
underbody may force the vehicle upwards. Alternatively, accelerated flows under the vehicle
may act to pull the vehicle downwards. Rocking of the vehicle on its suspension would present
an angled underbody to the flow. Differential water levels upstream and downstream of the
vehicle lead to an uneven buoyancy. The nature and magnitude of each of these factors will
vary in different flow depths, velocities and flow regimes. These aspects could not be quantified
in the present study.
Unsteady flows will rock the vehicle from side to side, affecting both hydrodynamic forces on the
vehicle (horizontal and vertical) and the instantaneous traction force available at the tyres. This
is particularly pertinent to vehicle stability due to the difficulty in regaining tyre traction once it is
lost.
While the additional complexity from variable flow conditions is acknowledged, this study only
considered the conservative case of steady flows.
5.1 Overview
The aim of the traction tests was to measure the force required to break traction at varying
water depths.
The friction of the tyres against the road is directly related to the weight at the tyre-road
interface and the coefficient of friction between the tyres and the road. As the water level rises,
the vehicle becomes more buoyant and less vertical force is available at the tyres to maintain
traction.
The testing procedure was designed to, as closely as possible, mimic the conditions of a vehicle
entering a flood flow. By flooding the test tank quickly to ensure minimal water ingress into the
vehicle cabin, and then towing the vehicle sideways (perpendicular to the direction of travel)
using a winch assembly attached to the wheel hub, it was possible to measure the force required
to break traction for the entire vehicle or specifically at the rear axle.
A rapid flood situation is typical of rising flood waters or driving into floodwaters, where the
ingress of water into the vehicle is limited by the effective cabin and door seals found in modern
vehicles. This situation when the vehicle is at its most buoyant, and therefore most vulnerable,
is considered the worst-case scenario. A vehicle submerged for some time will be filled with
water, be less buoyant and therefore require more force to move.
The winch test assembly consisted of a vertically arranged vehicle winch, a dynamometer (to
measure the forces), and a pulley. This allowed the winch and instrumentation to be clear of the
water during testing (see Figure 4-1).
Recent research by Haynes et al., (2016) noted that vehicle related flood fatalities in 4WDs have
increased in the last 15 years in Australia. Testing of the Nissan Patrol 4WD provided further
information for vehicles that may be more likely to be deliberately driven into floodwaters. This
test case is also relevant to flood rescue vehicles, which are generally by 4WD vehicles or larger.
Photographs of the test vehicles are provided in Figure 5-1 and Figure 5-2.
Figure 5-1: Toyota Yaris in WRLs test facility, preparing for a winch test from both axles
Figure 5-2: Nissan Patrol in WRLs wave facility, preparing for a winch test from the rear axle
The Toyota Yaris was tested in WRLs tow tank. The vehicle weight was estimated using the
specified vehicle wet weight (1045 kg) plus the driver load of 60 kg (chosen for a small to
medium sized adult). The tow force is the average of two tests conducted on the vehicle towed
from both axles on a wet concrete floor.
The Nissan Patrol 4WD coefficient of friction was tested twice, on both a wet floor and dry floor.
Tests of the coefficient of friction were conducted by towing the vehicle from the rear axle only.
The weight of the 4WD vehicle at the rear axle was measured directly by using the dynamometer
to lift the rear wheel vertically. The wheel was lifted only enough for the tyre to clear the
ground, to minimise unloading of the other tyres. This approach may overestimate the vehicle
weight, resulting in an underestimate of the coefficient of friction.
The results of these tests are provided in Table 5-2. A comparison between the two conditions
for the Nissan Patrol is consistent with Wong (1993) which indicates that there is little difference
in the coefficient of friction on wet or dry concrete.
In all cases the vehicle was winched sideways (the winching cable was perpendicular to the
driving direction). The cable arrangement was set so that the cable was perpendicular to the
ground and no vertical loading was applied by the winch. Tyres were checked daily to ensure
the specified tyre pressure was used throughout the testing program. The dynamometer was
zeroed immediately prior to each test. The output from the dynamometer was logged at 5Hz for
data post-processing. The accuracy of the dynamometer is 8 kg, being 0.1% of its full scale 8
tonnes, and has a measurement error of +/-2 kg.
The resulting time series of traction force as measured by the dynamometer were subsequently
analysed. The peak traction reading for each test was logged and was used for the analysis of
vehicle stability later in this study.
The Toyota Yaris testing was conducted for three (3) scenarios:
With an equivalent driver occupant only of weight 60 kg (small adult), winched from the
rear axle.
With a driver occupant only of weight 60kg, winched from both axles.
With four occupants having a combined weight of 240 kg distributed between 4 seats,
winched from the rear axle.
The rear axle only tests provided the stability limit state as it is expected that the reduced
vehicle weight over this axle will lose traction first (see discussion in Section 2). The tests on
winching the whole vehicle (both axles) provided data for the calculation of the coefficient of
friction, as well as validation data for assessment of the vehicle stability as a whole. To ensure
the minimum amount of water ingress into the Toyota Yaris, the basin was rapidly filled in less
than two minutes from an adjacent reservoir basin using a flood gate (Figure 5-4).
The Nissan Patrol 4WD, was tested in two (2) WRL facilities. Initial tests in the first facility
(wave basin) provided water depths up to 660 mm. Subsequent modelling in the second facility
provided deeper water conditions (of up to 1.3 m) which enabled the 4WD vehicle to float with
the rear wheels lifted off the ground. All 4WD testing was conducted with loading for a single
driver occupant (60kg), and was winched from the rear axle only.
Due to the slower fill time of the second laboratory facility (Lab 2), accessible leakage points in
the 4WD cabin were sealed with silicon, and a submersible pump was used to evacuate water
ingress. In this way the vehicle would remain buoyant for an indefinite period, even at high
water levels.
For both test methods, there was some small pooling of water towards the front of the vehicle
cabin, estimated to be less than 50L in both cases. Pooling in the floor of the cabin was always
at the front of the vehicle because the higher rate of buoyancy at the rear of the vehicle tilted
the vehicle forwards. This filling would increase the resulting measured traction force somewhat.
However, it is considered negligible because the majority of the effect would be at the front
wheels, while the rear is the critical point of vehicle instability. Some ingress of water would be
anticipated into vehicles in real world flood emergencies in all but the most extreme rapid
submersion scenarios (e.g. driving at speed into rapidly deepening waters).
Figure 5-6 The lead author demonstrating zero traction at the rear wheels for the Nissan Patrol
4WD in WRLs Lab#2 testing tank. Water depth is approximately 1.0m.
Results of the peak measured traction force (the maximum force measured) for tests on the
three (3) cases for the Toyota Yaris are provided in Table 5-3. Results of the peak measured
traction force for tests on the Nissan Patrol 4WD are provided in Table 5-4.
Measurement Uncertainty
The accuracy of the dynamometer is 0.1% of full deflection (8kg or 78.48N). Other sources of
error, such as pulley friction, are estimated to be less than 100N. Error in the measurement of
water depth is largely due to water level fluctuations following rapid filling and are estimated to
be less than 20 mm.
Figure 5-7: Traction force time series for tests on the Toyota Yaris with one driver occupant,
winched from the rear axle only
Figure 5-9: Traction force time series for tests on the Toyota Yaris with one driver occupant,
winched from both axles
Table 5-3: Measured peak horizontal traction force results for the Toyota Yaris
One Occupant, Rear Axle One Occupant, Both Axles Four Occupants, Rear Axle
Water Water Water
Depth (m) Horizontal Force Depth D Horizontal Force Depth D Horizontal Force
D (kN) (m) (kN) (m) (kN)
0.00 4.670 0.00 10.202 0.00 7.083
0.00 4.728 0.00 9.653 0.00 7.828
0.00 4.513 0.10 9.398 0.07 6.200
0.10 4.199 0.10 9.947 0.15 6.396
0.16 4.258 0.10 9.280 0.15 6.377
0.16 4.925 0.15 9.143 0.20 6.789
0.19 4.081 0.15 9.849 0.30 5.847
0.20 3.885 0.19 7.770 0.42 4.885
0.25 3.983 0.21 9.575 0.52 3.983
0.28 4.218 0.30 7.848 0.57 3.178
0.35 4.042 0.30 7.495
0.40 2.786 0.40 6.239
0.49 1.373 0.40 7.652
0.50 2.021 0.41 5.886
0.56 0.000 0.48 5.454
0.60 0.000 0.50 6.710
0.61 0.000 0.52 5.808
Two passenger scenarios i) driver only; and ii) driver plus three (3) passengers were tested in
the Toyota Yaris. These results showed that the number of passengers in the car affected the
vehicle stability threshold and that the limiting case was the case with the driver only. The
additional passengers increased the overall vehicle weight resulting in an increase in the traction
force. Tests were also conducted to determine the difference between the threshold force to
move sideways by pulling both front and rear axles and the rear axle only. Test results noted
that, as is observed in video footage of real flood situations, the limiting case is the case of rear
axle instability where the rear (no engine) end of the car floats first causing the car to pivot on
the front wheels an roll in the direction of flow. Since the limiting case was shown to be rear
axle instability, further testing of the Nissan Patrol 4WD was for the rear axle only.
Figure 5-12: Measured peak horizontal traction force for a Nissan Patrol 4WD 1998, winched from
the rear axle only, with one occupant
6.1 Overview
Hydrodynamic force testing was conducted on a scale model vehicle to determine the coefficient
of drag (CD) for the vehicle by direct measurement of hydrodynamic drag forces on the vehicle.
The tested flows were typical of those capable of destabilising a vehicle and ranged from
subcritical (Fr = 0.3) to supercritical (Fr = 4.2) flows. The resulting coefficient of drag
determined in the testing was used in Equation ( 4 ) to develop the stability curves presented in
Section 7.
Hydrodynamic testing was conducted in a 1 m wide tilting flume. The flume was a rectangular
channel with a parabolic entrance condition providing smooth and calm inflows. The tail water
condition could be controlled with a hinged gate at the tail of the flume. In all tests the flume
was kept horizontal to ensure the horizontal force component was accurately measured.
The model scale was chosen to suit available diecast model vehicles. A scale of 1:18 is
commonly used for scale cars and provides very good replication of body dimensions and detail,
including underbody roughness. Unfortunately a model identical to the vehicle used in the winch
tests could not be sourced. Instead a Toyota Yaris 2005 Hatch (rather than a sedan) was used
in the testing. This provided similar body features, curvature and underbody roughness, though
the overall body length was different.
Model scaling was based on geometric similarity with an undistorted length scale of 1:18 used
for all tests. The scaling relationship between length and time was determined by Froude
similitude, with the following relevant scale ratios (prototype divided by model) being adopted
for the model:
Length ratio LR = 18
Time ratio TR = LR0.5 = 4.243
Velocity ratio VR = LR0.5 = 4.243
Discharge ratio QR = LR2.5 = 1374.62
Force ratio FR = LR3 = 5832
The model vehicle was suspended by fine wire cables to a rigid frame and an adjustment
mechanism that allowed the vehicle tyres to be held just above the surface of the flume. The
cables were attached to each of the four wheels. The vehicle was oriented so that the flow was
directly perpendicular to the vehicle (90 degrees to the direction of vehicle travel).
The weight of the model was such that the vehicle was never buoyant under any flow condition,
so that only the hydrodynamic forces were measured. Buoyancy, friction and other forces were
not measured in the model. The vehicle was held at a fixed level, such that loading and
unloading of the springs due to buoyancy or rolling of the vehicle from side impact forces was
not considered.
For the horizontal hydrodynamic force tests, fine wire cables were attached between the
upstream side wheels (front and rear) and load cells mounted on a rigid frame upstream of the
vehicle, see Figure 6-1. The load cells both constrained the vehicle in the longitudinal direction,
as well as measuring the horizontal hydrodynamic forces on the vehicle. Forces were logged at
20Hz.
Figure 6-1: Test assembly for testing of the horizontal hydrodynamic forces. Flow is from the left
of the picture
Instrumentation
Forces in the model were measured using Futek LSB210 submersible load cells with full scale
range of 10 Lbs (4.536 kg). A coupled amplifier was used, and sampled at 20 Hz with a National
Instruments acquisition card () and LabVIEW.
Flow rates were measured using an electromagnetic flow meter with pipe diameter 100 mm.
Flow depths were measured using a point gauge.
Figure 6-2: Testing of the model in subcritical flows: Fr = 0.3, d= 0.71m, v = 0.80m/s
(prototype)
The model was tested by suspending the vehicle and test rig over a broad-crested weir. The
weir was constructed using a single sheet of plywood, approximately 19 mm thick, on the bed of
the flume (Figure 6-3). No tail water control was used in these tests as flow over a broad
crested weir is designed to transition the flows from subcritical flows upstream of the crest to
supercritical flow downstream (i.e. the broad-crested weir acts as a flow control).
Note that it is difficult to determine the exact Froude number in these conditions. The flow over
the weir is undulating, so that the Froude number varies above and below unity with distance.
Since the width of the vehicle covers some distance it also covers a range of Froude numbers.
Supercritical flows were developed in the scale test facility using an undershot gate as presented
in Figure 6-4.
Coefficient of Drag
Testing showed a range in the coefficient of drag across all flow regimes from 1.0 to 1.8. Values
in the subcritical range were quite scattered, but showed a slight trend upwards over the range
of Froude number from 0.30 to 0.76.
Critical flows through the tested range of supercritical flows (up to Fr= 4.2), showed a smaller
range of coefficient of drag of 1.1 to 1.6. However, the lower values were measured in tests
where the flow depths were lower than the body of the vehicle, so impact is predominantly with
the wheels alone. Flows which impacted the body of the vehicle typically ranged from CD of 1.4
to 1.6.
Force Force
Depth Velocity
Test Rear All CD Fr
(m) (m/s)
(kN) (kN)
Subcritical 1 0.315 0.589 0.378 1.01 1.08 0.53
2 0.251 0.461 0.468 0.82 0.98 0.38
3 0.490 0.927 0.630 0.90 1.14 0.36
4 0.437 0.781 0.711 0.80 1.06 0.30
5 0.805 1.639 0.828 0.97 1.26 0.34
6 0.787 1.685 0.783 1.03 1.24 0.37
7 1.312 2.869 0.585 1.37 1.65 0.57
8 1.283 2.770 0.585 1.37 1.60 0.57
9 1.872 3.983 0.666 1.44 1.80 0.56
7.1 Observations
This research investigation has produced two sets of results that can be consider both separately
and in combination.
Figure 7-1: Vehicle stability deducted from prototype scale measurements rear wheel horizontal
force thresholds versus water depth
A salient observation from these test results is the clear distinction between the stability of the
vehicles when the water level is below the vehicles floor pan and once it reaches a point just
above the floor pan. The buoyancy force of the air trapped in the vehicle cabin is analogous to a
bubble floating on the water surface. The results in Figure 7-1 show that as soon as floodwaters
are deeper than the floor of the vehicle and begin to engage with the air trapped in the vehicle
cabin, the force required to move the vehicle sideways rapidly decreases. For the Toyota Yaris,
this threshold depth for force reduction is approximately 0.2 m. As the flood water depth
increases above 0.2 m, the force required to move the Toyota Yaris sideways rapidly decreases
until the rear wheels are floating off the ground and effectively zero force is required to move
the car sideways at a depth of 0.6m.
Recall that a vehicle becomes unstable when the hydrodynamic force (FH) of the incident
floodwaters overcomes the friction force (FF) (reduced by vehicle buoyancy) i.e.:
Since FF has been measured in the prototype vehicle testing and the flow depth and velocity
have been measured in the hydrodynamic scale model testing, the unknown parameter in the
hydrodynamic force equation the coefficient of drag (CD) can be estimated. Recalling that, the
hydrodynamics force can be expressed as:
A comparison of the hydrodynamic model results and the prototype test results in Figure 7-2
shows that the hydrodynamic testing reproduced a range of scaled force results that
encompassed the prototype stability force testing.
The hydrodynamic testing also provided interesting results and observations on flow dynamics as
the flood depth rises relative to the vehicle. It was observed that the hydrodynamic force on a
vehicle rapidly rises when the flow depth meets the vehicle floor as drag is felt not only at the
four wheels but across the whole side and undercarriage of the vehicle. At the same time as the
flow depth interacts with the floor of the car, the stabilising weight of the vehicle begins to
reduce the downward force on the tyres, and subsequently the traction available at the wheels,
as the vehicle becomes buoyant. Additionally, it is likely that some hydrodynamic uplift may be
felt on the upstream side of the vehicle due to the impact of water against the lower parts of the
car body. All these hydrodynamic phenomena confirm the significant reduction in vehicle
stability as flow depths meet the vehicle floor pan.
The information compiled in Table 7-1 demonstrates that there are a number of vehicles that are
lower, lighter and/or longer than the Toyota Yaris test vehicle, and so will be more vulnerable in
floods. As noted earlier, a significant factor for vehicle stability is the buoyant volume of air in
the vehicle cabin compared with the weight of the vehicle. Along with a trend towards lighter
bodies, contemporary passenger vehicles tend to have more spacious cabins, integration of
chassis and body components, and smaller engine bays. This is most notable in hatch body
vehicles, where the cabin extends to the car boot, and the engine bay is minimised. Similarly,
Vehicle body dimension limits in Australia are set by the Roads and Maritime Services (Vehicle
Standards Information No. 5, 2012). This limits vehicle body clearance to 100 mm minimum.
This is significantly lower than the vehicles tested in this study. After market modifications to
small vehicles often include lowering the vehicle and this is likely to further reduce the stability
of these vehicles compared to the Toyota Yaris as tested.
Vehicles with rear mounted engines will have a similar response to forward mounted engine
vehicles, except that the front wheels will lose traction first. Travelling into deepening waters in
a rear engine mounted vehicle with the vulnerable front wheels first, means that the vehicle may
well be destabilised earlier and therefore be more vulnerable than a vehicle with a front mounted
engine.
The Nissan Patrol 4WD as tested can be considered representative of larger 4WD vehicles.
There is a trend to soft 4WD vehicles and sport utility vehicles (SUVs) that may be capable of
light off-road activities, but will be substantially more vulnerable in floodwaters than the results
for the testing the Nissan Patrol would indicate since these vehicles have less mass and are
comparatively lower to the ground.
Measurement uncertainty at both prototype and model scale. This is a minor and
quantifiable uncertainty, but nevertheless present.
Scaling effects. While the scale adopted for the modelling (1:18) was suitable for the
major forces of gravity and inertia, minor forces acting on the vehicle may not be well
represented at this scale;
o The impact of very turbulent, pulsating or varying flows has not been considered
but is important in initially destabilising a vehicle;
o The changes to vehicle height and roll angle (loading and/or unloading of the
suspension) due to buoyancy and side impact forces were not considered in the
hydrodynamic tests;
o The testing was conducted in ideal conditions, with regular channel dimensions
and steady flows. This is rarely the case in real world conditions;
o Uplift forces on the vehicle were poorly determined in the testing, this is largely
due to the magnitude of surface tension forces compared to the weight of the
vehicle. In other words, when considering the vertical forces, the weight of the
water raised by surface tension was significant compared with the uplift forces. A
larger scale and careful handling of this is required.
The Coefficient of Friction between tyre and road surface in actual flood conditions, as
opposed to wet road testing, is not well understood.
Only two vehicles were tested and they do not represent the limiting case for vehicle
stability.
These uncertainties are largely unquantifiable and so a level of conservatism is required when
interpreting and applying the test data in this report.
The prototype scale testing of vehicles conducted for this investigation has provided some stark
information on the vulnerability of vehicles as they interact with floodwaters.
Testing showed that the stabilising force due to the weight of a vehicle begins to rapidly reduce
once floodwaters rise above the floor of the vehicle and the bubble of air contained in the
vehicles cabin imparts an uplifting buoyancy force on the vehicle. Test results showed that
when flow depths increase to moderate levels, even large, heavy vehicles will reach a limiting
depth where they completely float. In the testing conducted for this investigation, a large
Nissan Patrol 4WD with a kerb weight of greater than 2.4 tonne was shown to completely
float in a water depth of 0.95 m. Smaller vehicles like the Toyota Yaris with weight of about
1.0 tonne are far more vulnerable due to their lower kerb weight. The Toyota Yaris tested in
this investigation completely floated in water 0.6 m deep.
Results from this investigation have demonstrated that moving flood waters have significant
momentum and can impart large hydrodynamic forces to vehicles exposed to the flow. This is
not a surprising outcome. To provide some context, each one (1) metre cube of water (1 m
by 1 m by 1 m box) of water weighs one (1) tonne. If a small car like a Toyota Yaris, which is
In fast moving flows, which can be upwards of three (3) to four (4) metres per second when
flows pass over bridges and causeways, this means that the smallest vehicles (see Table 7-1) on
Australian roads can become vulnerable when the flow is above the vehicle floor level in flood
depths as shallow as 0.15 m (15 cm).
Many 4WD manufacturers highlight the ability of 4WD vehicles to operate in water. For
example, the Ford Ranger is advertised as being able to negotiate water up to depths of 0.8 m
(http://www.motoring.com.au/new-ranger-in-deep-water-24397/ accessed 21 March 2017).
The alarming aspect of this claim is that the testing conducted for this investigation showed that
in water 0.8m deep, a 4WD vehicle would become unstable and vulnerable to being washed off a
bridge or causeway in flow travelling at about walking pace (i.e. 2 m/s or 7.2 km/h).
Since the force of flowing water is proportional to the velocity squared (see equation 3), small
increases in flow speed result in significant increases in the force the flow can impart.
Observations of flood flows in the field note that they rarely pass over a roadway steadily.
Rather, observations of real floods show that flows can change rapidly increasing and decreasing
in both depth and speed in a surging or random pulsing manner. In an operational context,
these often rapidly changing flow conditions make it very difficult to determine whether a
floodplain crossing is safe or not.
For flow depths that impact the body of the vehicle, there is no simple, safe guidance for vehicle
stability. Flows impacting the vehicle body may be safe for low depths and low velocities, but it
is highly variable from vehicle to vehicle, and as noted above, judgement of flow velocity (and
often depth) is difficult in real world situations.
In a flood emergency, the best advice remains to simply avoid driving in floodwaters.
The testing conducted for this investigation provides the opportunity to review the flood hazard
curves for vehicles derived for floodplain planning and emergency management, such as the
hazard curves published in ARR P10 (Shand et al., 2011) and AEMI Handbook 7 (2014). These
hazard curves provide a means for floodplain and emergency managers to classify flood
behaviour, as described by measured or predicted flood depths and flood flow velocities, with a
danger rating. The hazard rating for vehicles has been in the past, been quantified by the best
estimate of flow conditions that would cause a vehicle to become unstable in floodwaters and
wash away downstream. An example of a set of hazard curves from AEMI Handbook 7 (2014) is
provided in this report as Figure 2-1.
Consideration of the prototype instability force testing and model scale hydrodynamic testing in
combination enables flow instability threshold conditions for vehicles to be developed and
compared to existing flood hazard curves. This assessment requires estimation of suitable values
for coefficients in the governing equations described in Section 4.2. In particular, the analysis
A conservative effective coefficient of friction of 0.3 was adopted on the basis of the discussion in
Section 4.2.1 and the ratios summarised in Table 5-2. This value was adopted using similar
reasoning to the findings of Bonham and Hattersly (1967) who considered a braking force
coefficient of 0.5 on a worst case scenario of a smooth causeway surface on a wet country
road with adjustments made for a reduced sideways force compared with breaking force (10%),
the reduced slipping force compared with peak breaking force (20%), and a further reduction
due to debris that may become caught under the wheels (20%). This results in an adopted
coefficient of friction of 0.3.
An adopted coefficient of drag of 2.0 provides a conservative value based on the range of
measured values provided in Section 6 as summarised in Table 6-1. Uplift is assumed to be
zero, but can be considered to be accounted for in the conservative value of coefficient of drag.
A comparison of data derived from the testing on the basis of these adopted coefficients with the
currently adopted curves from ARR P10 is provided in Figure 7-3 for the Toyota Yaris
(representative of small cars) and in Figure 7-4 for the Nissan Patrol (representative of large
4WD vehicles.
Figure 7-3: Stability Curve for a Toyota Yaris based on full scale traction tests, upper limits for
the Coefficient of Drag CD = 2.0 based on scale testing, and points scaled to a flood condition
Coefficient of Friction of 0.3
At face value, the test results presented in Figure 7-3 indicate that the Toyota Yaris is more
stable than the interim vulnerability threshold condition for a small passenger vehicle from ARR
P10 expressed in terms of a flow velocity times depth product (v.D). The interim adopted
Similarly, the test results presented in Figure 7-4 indicate that the Nissan Patrol 4WD is
significantly more stable than the interim vulnerability threshold for a large 4WD vehicle from
ARR P10 of 0.5. The hazard threshold estimated from the test results for the Nissan Patrol has a
factor of safety of more than two (2) times the ARR P10 interim hazard guidelines.
Figure 7-4: Stability Curve for a Nissan Patrol based on full scale traction tests, upper limits for
the Coefficient of Drag CD = 2.0 based on scale testing, and points scaled to a flood condition
Coefficient of Friction of 0.3
There are, however, a range of limitations and in the testing regime which need to be carefully
considered when interpreting the test results from this investigation into a form suitable for
classifying flood hazard behaviour for floodplain planning or emergency planning and operations.
Firstly, when considering the vulnerability classification for small passenger vehicles, the Toyota
Yaris, while small, is far from the smallest, most vulnerable vehicle on Australian roads. Table
7-1 lists a number of vehicles that have lower road clearance and lighter kerb weight than the
Toyota Yaris.
Similarly, while the Nissan Patrol might be considered representative of large 4WD vehicles
there are numerous smaller 4WD vehicles that are lighter and have less ground clearance in this
class of vehicles. 4WD vehicles are also more likely to have larger tyres for off-road driving that
provided added buoyancy to the vehicles.
Secondly, there are a number of mitigating factors within the testing program that make it
sensible to apply a conservative interpretation of the results for a real-world application. These
The laboratory testing was conducted in ideal conditions, with regular channel
dimensions and steady flow. These conditions would rarely be encountered in real world
conditions. In the real world, flows are often sediment laden and sometimes contain
large debris both of which would increase the flow force on an exposed vehicle. Road
surfaces in flood affected areas are also often uneven with potholes and loose gravel
making full contact of all wheels on the submerged road surface problematic;
Laboratory testing was completed with stationary vehicles. In real flood conditions, most
cars will be driven into floodwaters and the act of rolling into the floodwaters may
contribute additional uplift forces and an associated reduction in friction not accounted
for in the testing;
The impact of turbulent, pulsating or varying flows has not been considered in the
testing but might often be prevalent in real world conditions, especially in flash flood
conditions, and is likely important in initially destabilising a vehicle;
The present test considered a simplified area for the drag force calculations. While the
flow would increase in the close proximity of the vehicle, in the present test the cross-
sectional area of the vehicle impacted by drag force was calculated based upon the flow
depth measured upstream of the vehicle. This would have resulted in a proportionally
larger flow velocity in the model and a reduced flow depth close to the vehicle;.
The changes to vehicle height and roll angle (loading and/or unloading of the
suspension) due to buoyancy and side impact forces were not considered in the scale
hydrodynamic tests;
Uplift forces on the model scale stationary vehicle were not well represented in the
testing, largely due to the magnitude of surface tension forces compared to the weight of
the vehicle. In other words, when considering the vertical forces, the weight of the
water raised by surface tension was significant compared with the uplift forces in the
model.
In due consideration of these aspects of the testing program i.e. i) that the test vehicles are not
representative of the most vulnerable vehicles in their class on Australian roads; and ii) that
there are limitations with the absolute accuracy of the results in the testing program, it is
appropriate that flood hazard thresholds for planning purposes remain conservative compared to
the test data presented here.
On this basis, the authors recommend that the interim flood hazard threshold curves from ARR
P10 summarised here as Table 7-2 for the two vehicle classes tested in this investigation be
adopted for use in future floodplain management and emergency management.
Limiting
Kerb Ground Limiting High
Class of Length Limiting Equation of
Weight Clearance Still Water Velocity
Vehicle (m) Velocity3 Stability
(kg) (m) Depth1 Flow
Depth2
Small
< 4.3 < 1250 < 0.12 0.3 0.15 3.0 D.V 0.3
passenger
Large 4WD > 4.5 > 2000 > 0.22 0.5 0.3 3.0 D.V 0.6
1 2 3
At velocity = 0 m/s; At velocity = 3 m/s; At low depth
A study of vehicle stability in floodwater flows was conducted at UNSW Sydney WRLs
laboratorys facilities incorporating full prototype scale testing of traction and buoyancy forces on
a vehicle in static water, and hydrodynamic force measurement on 1:18 scale model vehicles.
The traction force tests conducted are novel in research on stability of vehicles, and show that
traction decreases rapidly with depths above the vehicles floor pan level. The testing confirmed
that the rear wheels (wheels not weighted by the engine) lose traction far earlier than the front
wheels.
Hydrodynamic testing showed that coefficients of drag for the vehicle in free surface water flows,
perpendicular to the vehicle, are far higher than typical wind tunnel based aerodynamic drag
coefficients. These are consistent with drag coefficients predicted by Hoerner (1965) for water
surface piercing drag and complicated by the vehicle geometry and low depth to vehicle length
ratios. The coefficient of drag varies non-linearly with the flow condition, but is generally within
the range 1.2 and 2.0 for typical, dangerous conditions.
The test program and analysis presents a novel approach to the derivation of stability curves for
vehicles in flood flows, by dissecting the forces that impact the vehicle, applying appropriate
factors of safety, and evaluating vehicle stability directly from the stability balance equation.
Observations from the testing indicate that where floodwaters impact the body of the vehicle,
the vehicle is at serious risk of losing traction and being washed downstream. Flows that do not
impact the body of the vehicle (hitting the wheels only) are generally safe for that vehicle.
Analysis of results from this investigation showed that the subject test vehicles were more stable
in flood flows than the flood hazard vulnerability thresholds typically applied for floodplain
management and emergency management as documented by guidelines such as Australian
Rainfall and Runoff Review Project 10 (ARR P10) (Shand et al., 2011) and AEMI Handbook 7
(2014).
However, mitigating factors described in this report, notably i) that the tested vehicles, while
representative, were not the smallest in their vehicle class and ii) that controlled laboratory
conditions need to be interpreted to uncontrolled real life flood conditions means that application
of a level of conservatism on this studys results is warranted.
8.1 Recommendations
In the course of this investigation, numerous issues were encountered, which with the fullness of
time and available funding, warrant further analysis.
1. Vehicles should primarily be tested with the vehicle oriented at or near perpendicular to
the flow. Consideration for vehicles at other angles may be beneficial, but it is expected
that the limiting case for vehicle vulnerability in floods is perpendicular to the flow;
2. Further research is required on the range of coefficients of drag for a range of vehicles in
free surface water flow, preferably at prototype scale;
4. One of the significant unknowns in vehicle stability is the coefficient of friction between
tyres and the submerged road in flood conditions. While the coefficient of friction
adopted in this study and others ( = 0.3) is conservative, a better understanding of the
effect of various surface types and the effect of debris on the road on tyre friction would
improve the accuracy of this studys outcomes.
5. Lifting of the vehicle (unloading the suspension) due to buoyancy was not considered in
the scaled hydrodynamic tests. Similarly, the rolling of the vehicle on the suspension
due to the side impact of the water was not considered in the scaled hydrodynamic tests.
Both of these would have further impacts on both horizontal and vertical hydrodynamic
forces on the vehicle. It is expected that a significantly larger scale would be required to
correctly model these forces.
6. Testing of the model vehicle was conducted on a broad crested weir, resembling a typical
road embankment, but only in a single location. It is recommended that testing be
conducted across a range of distances downstream of the weirs upstream edge.
The authors of this report acknowledge the contribution of the NSW State Emergency Service
and NSW Office of Environment and Heritage for their substantial funding of this investigation.
This research could not have progressed without the generosity of NRMA Insurance who
provided the test vehicles at no cost to the investigation.
AEMI. (2014) Australian Emergency Management Handbook: Managing the floodplain: best
practice in flood risk management in Australia, (AEMI Handbook 7, 2014)
Bird, G. and Scott W.J.O. (1936) Studies in Road Friction, I. Road Surface Resistance to Skidding
D.S.I.R., Road Research Tech. Pap. No.1 H.M.S.O., 1936
Bird, G. and Scott W.J.O. (1936) Studies in Road Friction, II. An Analysis of the Factors Affecting
Measurement D.S.I.R., Road Research Tech. Pap. No.2 H.M.S.O., 1936
Bonham, A.J. and Hattersley, R.T. (1967) Low Level Causeways, University of New South Wales,
Water Research Laboratory, Technical Report No. 100, August 1967
Bowen, W and Alexander, K. (2011) Surface Piercing Drag in CFD. In: Proceedings of the ASME
30th International Conference on Ocean, Offshore and Arctic Engineering. OMAE2011-49651
Cox, R.J., Shand, T.D. and Blacka, M.J. (2010) Appropriate Safety Criteria for People in Floods.
WRL Research Report 240. ARR Project No. 10, prepared for Institution of Engineers Australia.
22p
DIPNR (New South Wales Government Department of Infrastructure, Planning and Natural
Resources) 2005, Floodplain development manual: the management of flood liable land,
Department of Infrastructure, Planning and Natural Resources, Sydney
Gerard, M. (2006). Tyre-road friction estimation using slip-based observers. Masters thesis.
Dept. Automatic Control, Lund University, Lund Sweden
Gordon, A.D. and Stone, P.B. (1973) Car Stability on Road Causeways. WRL Technical Report
No. 73/12. 5p + Appendices
Haynes, K., Coates, L., Dimer de Oliveira, F., Gissing, A., Bird, D., van den Honert, R., Radford,
D., DArcy, R, Smith, C. (2016) An analysis of human fatalities from floods in Australia 1900-
2015. Report for the Bushfire and Natural Hazards CRC
Haynes, K. and Gissing, A. (2016) Flood deaths are avoidable: dont go in the water. Macquarie
University. https://theconversation.com/flood-deaths-are-avoidable-dont-go-in-the-water-60615
published June 7, 2016 accessed 27 September 2016
Hoerner, S.F., (1965) Fluid-dynamic drag: practical information on aerodynamic drag and
hydrodynamic resistance. Midland Park, N.J.: Dr.-Ing. S.F. Hoerner
International Federation of Red Cross and Red Crescent Societies (2015) World disasters report.
Geneva, Switzerland
Jonkman, S.N. (2005) Global perspectives on loss of human life caused by floods. Natural Hazards,
34 (2), pp.151-175
Keller, R.J., and Mitsch, B. F., (1992) Stability of Cars and Children in Flooded Streets In:
International Symposium on Urban Stormwater Management (1992 : Sydney, N.S.W.).
International Symposium on Urban Stormwater Management: Preprints of Papers. Barton, ACT:
Institution of Engineers, Australia, 1992: 264-268. National conference publication (Institution of
Engineers, Australia) ; no. 92/1. ISBN: 0858255472
Martnez-Gomariz, E., Gmez, M., Russo, B. and Djordjevi, S. (2016) Stability criteria for
flooded vehicles: a state-of-the-art review. J Flood Risk Management. doi:10.1111/jfr3.12262
Munson, B. R., Okiishi, T. H., Huebsch W. W. and Rothmayer A. P. (2013) Fundamentals of Fluid
Mechanics. 7th edition
NSW SES (2016) Flood safety warning following 300 flood rescues. Web article:
http://www.ses.nsw.gov.au/news/76365/77179 accessed 27 September 2016
Oshikawa H., Oshima T. and Komatsu T. Study on the Risk for Vehicular Traffic in a Flood
Situation (in Japanese). Adv River Eng 2011, 17, 461466
Roads and Maritime Services (2012) Vehicle Standards Information No. 5 Rev6 VSI No.5
Shand, T.D., Cox, R.J., Blacka, M.J. and Smith, G.P. (2011) Australian Rainfall and Runoff
Revision Project 10: Appropriate Safety Criteria for Vehicles - Literature Review. Australian
Rainfall and Runoff Revision Project 10. Stage 2 Report. Prepared by the Water Research
Laboratory. P10/S2/020. February 2011
Smith, G.P., Davey, E.K., and Cox, R.J. (2014) Flood Hazard. UNSW Australia Water Research
Laboratory Technical Report 2014/07
Shu, C., Xia, J., Falconer, R.A. and Lin, B. (2011) "Incipient Velocity for Partially Submerged
Vehicles in Floodwaters". Journal of Hydraulic Research, 49 (6), pp.709-717
Toda K., Ishigaki T. and Ozaki T. (2012) Experiment study on floating car in flooding.
International Conference on Flood Resilience: Experiences in Asia and Europe, Exeter, UK, 2013,
6.
Xia, J., Teo, F.Y., Lin, B. and Falconer, R.A. (2011) "Formula of Incipient Velocity for Flooded
Vehicles". Natural Hazards, 58 (1), pp.1-14
Xia J., Falconer R.A., Wang Y. and Xiao X. (2011) New criterion for the stability of a human
body in floodwaters. J Hydraul Res 2014, 52, 93104. doi: 10.1080/00221686.2013.875073
Xia J., Falconer R.A., Xiao X. and Wang Y. (2013) Criterion of vehicle stability in floodwaters
based on theoretical and experimental studies. Nat Hazards 2013, 70, 16191630. doi:
10.1007/ s11069-013-0889-2.